Розробка моделей навчання на основі машинного навчання з квантовим відпалом для оптимізації навчання в цифрову еру

The object of this study is the prediction of digital learning achievement. The problems solved in this study are the low accuracy and efficiency of the prediction model caused by the complexity of the learning data and the limitations of conventional tuning methods such as grid search and random search which are unable to optimally navigate the wide and non-linear parameter space. The results obtained show that the integration of quantum annealing into the hyperparameter optimization process can significantly improve model performance. Model accuracy increased from 82% to 91%, with consistent improvements in precision, recall, and F1-score. The model also showed faster convergence and lower losses on both training and testing data, indicating better generalization capabilities to new data. Interpretation of these results concludes that quantum annealing can navigate the parameter space efficiently, exploring combinations of values that are unreachable by conventional methods. The main feature and characteristic of these results lies in its ability to combine the computational efficiency of LightGBM with the exploration of complex solutions through quantum methods, making it very suitable for dynamic learning problems. The scope and conditions of practical use of the developed model include digital-based learning management systems, adaptive learning platforms. These findings are relevant to be applied in the development of artificial intelligence-based education systems that support personalization in the current era of digital transformation